Artificial Intelligence Data Science

Dynamic pricing and bundling in real time

The basis of current software solutions for dynamic pricing is the combination of different pricing strategies and the creation of a (more or less) complex set of rules:

  • Time-dependent: The time of day but also seasonal conditions are taken into account.
  • Peak Load Pricing: As demand increases, prices pick up. The trader can also react in reverse order to place his prices below those of the competition. Recognize their own systems that the competition can no longer deliver, prices are raised significantly.
  • Penetration strategy: The conscious decision to set prices against the competition can also be part of a penetration strategy. It is about always offering the lowest price to quickly gain market share. Once this is achieved, the prices will be tightened again.

Around the basic price strategy then further rules and filters can be activated and adjusted. Classic example: The use of an Apple device suggests more purchasing power. So here can be tried to enforce a higher price.

By combining different rules, it is possible to design complex structures for pricing. However, tools for dynamic pricing are purely reactive systems at this stage of development. If one or more conditions meet, this leads to a price adjustment.

In order to act proactively, data and information from various areas can also be consulted and interpreted through the use of machine learning. This includes purchase histories, sales figures and external data on products and the market. The underlying learning algorithm is now constantly exploring the interaction between prices and sales. The hypothetical optimal price is then confirmed or refuted on the fly using A / B testing.